IEEE Solid-States Circuits Magazine - Summer 2022 - 17

That, of course, is a hideous oversimplification,
but it's good enough
to understand what's going on here.
You can find cross-sections of bipolar
transistors structures most anywhere,
but if you have a copy of [8],
check out figures 2.15 and 2.33.)
Unlike 2D integration,
though,
merged function or superintegration
doesn't have a clear path forward.
Clever designers are using it
here and there to save a little real
estate, but it's really just a collection
of tricks. It's opportunistic with no
methodology per se. That makes it
sound like a dead end, but this does
not deter the engineers of 1981,
probably because the same thing is
being said of all analog design: it's
just a collection of tricks.
One instance of superintegration
that does enjoy wide commercial
success is " integrated injection
logic, " known fondly as I-squared-L.
The combination of lateral PNP
action with multicollector NPNs on a
common base makes for extremely
dense logic in a conventional bipolar
process. It is used for all kinds
of " mixed-mode " ICs in the early
1980s,
including ADCs,
instrument
chips, and even a digital watch. So,
in 1981, there is reason to hope
that merged-function design has a
bright future.
But, the most surprising thing
about the IC business in 1981 is
the idea that the circuit designer
can-and should-mess with transistor
design. We're not just talking
about length and width or number
of stripes. Even the shape and overlap
of various diffusions is tweakable.
Of course, there are spacing
rules that must be observed and a
well-known set of best practices,
but if you have a good reason to get
creative, it is allowed. In the future,
of course, the FinFET will close and
lock this door, but in 1981, it's open
season, baby!
Meanwhile, Back in the Lab . . .
The germ of an idea has come to
our grad student from two industry
legends. Barrie Gilbert has figured
out how to steer the active area of
a distributed device [5], but to date,
it is largely an answer in search of
a question. Paul Brokaw reckons
that such functionality ought to be
useful for analog-to-digital
conversion but doesn't
know exactly how. This all
seems like it should be fertile
thesis territory, but
so far, it's proven barren
and rocky.
The breakthrough,
when it comes, seems
too good to be true. A
variation on Gilbert's
idea catapults Brokaw's
ADC into reality. It is anything
but pretty (see Figure 1), and it takes
a bit of explaining to make anyone
believe it.
Meet the Wiggler
You see, our grad student is on the
trail of something Brokaw calls a
" bit-slice " ADC, shown in Figure 2.
The concept comes from the recognition
that each individual output
bit of an ADC toggles periodically as
the input of the ADC is swept across
the full-scale range. If you can come
up with a circuit whose output
While there is
much excitement
about everhigher
levels of
integration, the
deep thinkers
among us
wonder if maybe
we've missed
something.
varies periodically with the input
signal, you can use that circuit to
generate the output bit directly. A
collection of such circuits, with different
periods, operating
in parallel, could
then create all of the
bits of the ADC output
code directly . . . and
very quickly.
Don't get confused by
the diagram in Figure 2:
this isn't about oscillators.
Each bit-slice block must
have a periodic response
to its input, not a periodic
response in time. Also, the picture
shows the outputs making a binary
pattern because that's the familiar
ADC output, but that really isn't
practical. Instead, the outputs form
a Gray code (see " Sixteen Shades of
Gray " ) so that the ADC isn't crippled
by bit transition mismatches.
Circuits that have a periodic
response to their input are unusual
but not unheard of. Brokaw's favorite
solution is shown in Figure 3 [3], [4].
Because of the frisky back-and-forth
response of the outputs to a swept
input, Brokaw christens the device
Out
In
Out
In
Out
In
Out
In
Subconverter
FIGURE 2: The bit-slice ADC architecture (binary coding).
IEEE SOLID-STATE CIRCUITS MAGAZINE
SUMMER 2022
17
Input
Output Word
IEEE Solid-States Circuits Magazine - Summer 2022

Table of Contents for the Digital Edition of IEEE Solid-States Circuits Magazine - Summer 2022
